U.S. patent number 7,881,191 [Application Number 10/579,170] was granted by the patent office on 2011-02-01 for method and apparatus for downlink multi-channel packet combined scheduling in mobile communication system.
This patent grant is currently assigned to Utstarcom (China) Co. Ltd.. Invention is credited to Sheng Liu, Baijun Zhao.
United States Patent |
7,881,191 |
Liu , et al. |
February 1, 2011 |
Method and apparatus for downlink multi-channel packet combined
scheduling in mobile communication system
Abstract
The present invention sets forth a method for performing packet
combined scheduling of dedicated transport channels for packet
services in UMTS downlinks, comprising the following steps: a)
prior to each DCH scheduling period, performing pre-selection
processing of a transport format combination of each DPCH according
to the predetermined restriction conditions for the DCH combined
packet scheduling, so as to determine a usable transport format
combination set for each DPCH; b) restricting a total downlink
transmit power of DCHs for NRT packet services to a schedulable
power not exceeding a schedulable power value in the estimation of
a total downlink power during said scheduling period; c) based on
the fairness of DCH transportation and the QoS requirements of the
DCH-borne services, determining weighted values which the
respective DCHs correspond to in the optimization of the DCH
combined packet scheduling; and d) based on the results of steps
a), b), and c), calculating the maximum number of bits which each
DCH is schedulable to output, using a 0-1 programming algorithm.
The present invention guarantees the fairness, priority and QoS
(Quality of Service) of different DCHs and can achieve maximum
total data throughput.
Inventors: |
Liu; Sheng (Guangdong,
CN), Zhao; Baijun (Guangdong, CN) |
Assignee: |
Utstarcom (China) Co. Ltd.
(Beijing, CN)
|
Family
ID: |
34578655 |
Appl.
No.: |
10/579,170 |
Filed: |
November 12, 2003 |
PCT
Filed: |
November 12, 2003 |
PCT No.: |
PCT/CN03/00952 |
371(c)(1),(2),(4) Date: |
January 24, 2007 |
PCT
Pub. No.: |
WO2005/048613 |
PCT
Pub. Date: |
May 26, 2005 |
Prior Publication Data
|
|
|
|
Document
Identifier |
Publication Date |
|
US 20090016275 A1 |
Jan 15, 2009 |
|
Current U.S.
Class: |
370/230; 455/449;
455/436; 370/233; 370/236 |
Current CPC
Class: |
H04W
72/1273 (20130101); H04W 28/06 (20130101); H04W
84/04 (20130101) |
Current International
Class: |
G01R
31/08 (20060101); G06F 11/00 (20060101) |
Field of
Search: |
;370/236,233
;455/436 |
References Cited
[Referenced By]
U.S. Patent Documents
Foreign Patent Documents
Other References
Chiussi, F.M ; Francini, A "Minimum-Delay Self-Clocked Fair
Queueing Algorithn for Packet-Switched Networks" IEEE vol. 3 Pub
1998. cited by examiner.
|
Primary Examiner: Harold; Jefferey F
Assistant Examiner: Harley; Jason
Attorney, Agent or Firm: Ladas & Parry LLP
Claims
What is claimed is:
1. A method for performing packet combined scheduling of dedicated
transport channels for packet services in UMTS downlinks, wherein
dedicated traffic channels (DTCHs) in logical channels are mapped
as dedicated channels (DCHs) in transport channels, and N DCHs, in
their respective input queues, queue up for being transported to
the corresponding M DPCHs, where M.ltoreq.N, the method being
characterized in that the method for performing packet combined
scheduling of the DCHs comprises the following steps: a) prior to
each DCH scheduling period, performing pre-selection processing of
a transport format combination of each DPCH according to the
predetermined restriction conditions for the DCH combined packet
scheduling, so as to determine a usable transport format
combination set TFCS.sub.m.sup.(2) for each DPCH; b) restricting a
total downlink transmit power .times. ##EQU00016## of DCHs for NRT
packet services to a schedulable power not exceeding a schedulable
power in the estimation of a total downlink power during said
scheduling period, i.e., the maximum allowable power value
P.sub.k.sup.Schedulaed, where P.sub.k,n denotes an average transmit
power required by the N-th DCH in the k-th scheduling period, and
P.sub.k.sup.Scheduled denotes the maximum allowable power allocated
to the DCHs which bear NRT packet services in the estimation of
downlink power in the k-th scheduling period, and calculating a
predicted/estimated value c.sub.k,n of a proportional coefficient
of an average transmit power P.sub.k,n of the n-th DCH and the
number of the bit(s) R.sub.k,n of the n-th DCH scheduled to be
output in the k-th scheduling period; c) based on fairness of DCH
transportation and QoS requirements of the DCH-borne services,
determining weighted values which the respective DCHs correspond to
in the optimization of the DCH combined packet scheduling; and d)
based on the results of steps a), b), and c), calculating the
maximum number of bit(s) which each DCH is schedulable to output,
using a 0-1 programming algorithm.
2. The method as defined in claim 1, wherein said determining the
usable transport format combination set TFCS.sub.m.sup.(2) for each
DPCH according to the predetermined restriction conditions for the
DCH combined packet scheduling in step a) further comprises:
placing a high-priority packet at the front of an input buffer
queue of a corresponding DCH; and performing the pre-selection
processing on TFCS based on the TFCS of the DPCH, the activated
DCH, and the length of the input queue of the DCH, thereby
obtaining the usable TFCS.sub.m.sup.(2) of each DPCH in the current
scheduling period.
3. The method as defined in claim 2, wherein the pre-selection
processing further comprises the steps of: i) selecting a TFC set
TFCS.sub.m.sup.(0)={(TF.sub.1, TF.sub.2, . . . ,
TF.sub.s.sub.m)}.sub.m.sup.(0) in which the size and number of
transport blocks which each DCH is scheduled to output meet the
limit of the TFCS which the DCH corresponds to, where m=1, 2, . . .
M, S.sub.m representing the dimension of the TFC of the m-th DPCH,
that is, S (m) DCHs are multiplexed on the m-th DPCH; ii)
determining whether each DCH in each DPCH has been activated in the
current scheduling period; if a certain DCH has been activated,
then removing all the transport formats combinations excluding the
transport format being currently used by the DCH from the usable
TFC set obtained by step (i) of the DPCH which the DCH corresponds
to, the usable TFC set of each DPCH obtained by step ii) being
marked as TFCS.sub.m.sup.(1)={(TF.sub.1, TF.sub.2, . . . ,
TF.sub.S.sub.m)}.sub.m.sup.(1), where m=1, 2, . . . M; and iii)
removing all transport formats combinations which satisfy the
following conditions from the usable TFC set of each DPCH obtained
by step (ii): the TFC contains at least one transport format
indicating that the number of the bit(s) transportable on the
corresponding DCH within the current scheduling period is greater
than the length of the current input buffer queue of the
corresponding DCH, the usable TFC set of each DPCH obtained by step
(iii) being marked as TFCS.sub.m.sup.(2)={(TF.sub.1, TF.sub.2, . .
. , TF.sub.S.sub.m)}.sub.m.sup.(2), where m=1, 2, . . . M.
4. The method as defined in claim 2, wherein the high-priority
packet is a status Protocol Data Unit packet of a Radio Link
Control layer of an Acknowledged Mode.
5. The method as defined in claim 1 wherein said method further
comprises, prior to the pre-selection processing in step a), a step
of performing DTCH scheduling allocation for the DCH by using the
Round-Robin, WFQ or WF.sup.2Q scheduling algorithm.
6. The method as defined in claim 1, wherein the proportional
coefficient c.sub.k,n in the current scheduling period is
predicted/estimated by using a linear prediction filter based on
the following equation:
c.sub.k,n=(1-.alpha.)c.sub.k-l,n+.alpha.c.sub.k-l,n, where .alpha.
is a filter coefficient and .alpha..epsilon.[0, 1]; c.sub.k-l,n is
a ratio of a measurement value of an actual transmit power of the
n-th DCH and the number of the bit(s) actually scheduled to be
output in the (k-l)-th scheduling period.
7. The method as defined in claim 1, wherein step c) further
comprises a step of determining the corresponding weighted values
of respective DCHs in the DCH combined packet scheduling
optimization based on a product of a dynamic weighted value
w.sub.k,n.sup.Fair and a static weighted value w.sub.n.sup.QoS,
wherein the static weighted value w.sub.n.sup.QoS is determined
based on the priorities of services borne by the n-th DCH; an
average scheduling index .eta..sub.k,n of the n-th DCH in the k-th
scheduling period is calculated based on the recorded scheduling
result of each DCH, and the dynamic weight value w.sub.k,n.sup.Fair
is determined based on the following formula:
.eta..times..times..eta. ##EQU00017##
8. The method as defined in claim 7, wherein the average scheduling
index .eta..sub.k,n is determined by one of the following methods:
i) smoothing filtering the scheduling index of the n-th DCH in the
past (k-l)-th scheduling period: .eta..times..times..eta.
##EQU00018## wherein L represents the number of the past scheduling
periods participating in the smoothing filtering; ii) adjusting the
formula in method i): .eta..times..times..lamda..times..eta.
##EQU00019## where factor .lamda. is .lamda..epsilon.(0, 1]; or
iii) performing the smoothing filtering by using a first-order
Infinite Impulse Response filter: .eta..sub.k,n=(1-.beta.)
.eta..sub.k-l,n+.beta..eta..sub.k-l,u, where filter coefficient
.beta. is .beta..epsilon.[0, 1].
9. The method as defined in claim 8, wherein the "scheduling index"
of the n-th DCH in the (k-l)-th scheduling period is defined as:
.eta..times..noteq..times. ##EQU00020## where, R.sub.k-l,n
represents the number of the bit(s) which the n-th DCH is scheduled
to output in the (k-l)-th scheduling period, and max{R.sub.k-l,n}
represents the maximum value in a discrete limited area of the
schedulable bits which the n-th DCH corresponds to in the (k-l)-th
scheduling period.
10. The method as defined in claim 1, wherein step d) further
comprises: converting the maximization of a target function in the
DCH combined packet scheduling into 0-1 programming minimizing the
target function by using the calculating results in steps a), b)
and c), then further converting the 0-1 programming into a linear
programming for processing, thereby calculating the number of the
bit(s) which each DCH is scheduled to output, wherein the target
function in the DCH combined packet scheduling is defined as:
.times..times. ##EQU00021## where R.sub.n should satisfy the
restriction condition: .times..times..ltoreq. ##EQU00022## and
R.sub.n.epsilon..PSI..sub.n, wherein .PSI..sub.n is a discrete
limited area; the target function minimized by the 0-1 programming
and converted from the target function in the DCH combined packet
scheduling is defined as: .times..times. ##EQU00023## the
restriction condition being .times.
.times..times..gtoreq..times..di-elect cons..times..times.
##EQU00024## where, S is a slack variable, and parameters W.sub.m,i
and C.sub.m,i in the above 0-1 programming are given in the
following formulas: .times..function. ##EQU00025##
.times..function. ##EQU00025.2## and based on formula
.times..times..function. ##EQU00026## calculating the number of the
bit(s) R.sub.j,i which each DCH is optimally scheduled to output,
wherein: q.sub.m,i represents a 0-1 indicator variable and
q.sub.m,i.epsilon.{0, 1}, i=1, 2, . . . , D.sub.m; D.sub.m is the
number of elements of the usable TFC set TFCS.sub.m.sup.(2) of
DPCH.sub.m; q.sub.m,i corresponds to each element in said TFC set,
respectively; DCHs multiplexed to the same DPCH are regarded as a
group, so M DPCHs have M groups of DCHs, each group containing
S.sub.m(m=1, 2, . . . M)DCHs, the DCHs being numbered as
DCH.sub.m,1, DCH.sub.m,2, . . . , DCH.sub.m,S.sub.m based on their
turns in the TFC of the corresponding DPCH.sub.m, and R.sub.n,
w.sub.n, c.sub.n are also correspondingly marked as R.sub.m,j,
w.sub.m,j, c.sub.m,j, where j=1, 2, . . . , S.sub.m, r.sub.m,j(i)
being the number of transmissible bit(s) in the current scheduling
period indicated by the transport format of the j-th DCH contained
in the l-th TFC in the TFCS.sub.m.sup.(2).
11. The method as defined in claim 1, wherein steps a), b) and c)
are executed in a parallel way.
12. The method as defined in claim 1, wherein the scheduling period
is the frame length of a physical channel.
13. An apparatus for performing packet combined scheduling of
dedicated transport channels for packet services in UMTS downlinks,
wherein dedicated traffic channels (DTCHs) in the logical channels
are mapped as dedicated channels (DCHs) in the transport channels,
and N DCHs, in their respective input queues, queue up for being
transported to the corresponding M DPCHs, where M.ltoreq.N; the
apparatus being characterized in that the apparatus for performing
the packet combined scheduling of the DCHs comprises: a
pre-selection processing unit for, prior to each DCH scheduling
period, performing pre-selection processing of a transport format
combination of each DPCH according to the predetermined restriction
conditions for the DCH combined packet scheduling, so as to
determine a usable transport format combination set
TFCS.sub.m.sup.(2) for each DPCH; a power restriction proportional
coefficient calculating unit for restricting a total downlink
transmit power .times..times. ##EQU00027## of DCHs for NRT packet
services to a schedulable power not exceeding a schedulable power
in the estimation of a total downlink power during said scheduling
period, i.e., the maximum allowable power value
P.sub.k.sup.Scheduled, where P.sub.k,n denotes an average transmit
power required by the N-th DCH In the k-th scheduling period, and
P.sub.k.sup.Scheduled denotes the maximum allowable power allocated
to the DCHs which bear NRT packet services in the estimation of
downlink power in the k-th scheduling period, and for calculating a
predicted/estimated value c.sub.k,n of a proportional coefficient
of an average transmit power P.sub.k,n of the n-th DCH within the
k-th scheduling period and the number of bit(s) R.sub.k,n which the
n-th DCH is schedulable to output within the k-th scheduling
period; a target function weighted value calculating unit for,
based on fairness of DCH transportation and QoS requirements of the
DCH-borne services, calculating corresponding weighted values of
respective DCHs in optimization of the DCH combined packet
scheduling; and a 0-1 programming-based optimum packet scheduling
calculating unit for, based on output results of the pre-selection
processing unit, the power calculating unit and the target function
weighted value calculating unit, calculating the maximum number of
the bit(s) which each corresponding DCH is schedulable to output,
by using a 0-1 programming algorithm.
14. The apparatus as defined in claim 13, wherein the pre-selection
processing unit further comprises: a DTCH scheduling module for
executing a DTCH scheduled allocation of the DCH by using
Round-Robin, WFQ or WF.sup.2Q scheduling algorithm; a priority
queuing module for putting a high-priority packets, such as a
status PDU of a RLC of the AM, at the front of the input buffer
queue of the corresponding DCH; a TFCS pre-selection processing
module for conducting the pre-selection processing on TFCS based on
the TFCS of the DPCH, the activated DCH, and the length of the
input queue of the DCH, thereby to obtain the usable
TFCS.sub.m.sup.(2) of each DPCH in the current scheduling
period.
15. The method as defined in claim 13, wherein said power
calculating unit further comprises: a linear prediction filter for
estimating a power restriction proportional coefficient of each DCH
in the current scheduling period based on a ratio of a measurement
value of each DCH's actual transmit power to the recorded actual
number of the bit(s) scheduled to be output in the previous
scheduling period.
16. The apparatus as defined in claim 13, characterized in that
said target function weighted value calculating unit further
comprises: a dynamic weighted value calculating module for
calculating the dynamic weighted value in the target function
according to the recorded scheduling result of each DCH based on
the fairness requirement of each DCH; a static weighted value
calculating module for determining the static weighted value based
on the priorities of services borne by each DCH; and a multiplier
for multiplying the dynamic weighted value and the static weighted
value of each DCH thereby to obtain the weighted value of the
target function.
17. The apparatus as defined in claim 13, characterized in that
said 0-1 programming-based optimum packet scheduling calculating
unit further comprises: a 0-1 programming parameter calculating
module for converting, based on the output results of the
pre-selection processing unit, the power restriction proportional
coefficient calculating unit and the target function weighted value
calculating unit, the maximization of a target function in the DCH
combined packet scheduling into 0-1 programming minimizing the
target function and calculating parameters W.sub.m,i and C.sub.m,i
in the 0-1 programming problem; a linear programming calculating
module for calculating the optimum solution vector of indicator
variables in the 0-1 programming problem of downlink DCH combined
packet scheduling by utilizing the parameters W.sub.m,i, C.sub.m,i
and the schedulable power estimation in the current scheduling
period, wherein the schedulable power estimation in the current
scheduling period is provided by a downlink power allocating unit
of a cell radio resource management module; and a DCH scheduled
output bit calculating module for calculating the number of the
bit(s) each DCH scheduled to be output in the current scheduling
period based on the optimum solution vector of the indicator
variables output by the linear programming calculating module.
18. The apparatus as defined in claim 13, wherein said
pre-selection processing unit, said power calculating unit, and
said target function weighted value calculating unit are operated
in a parallel way.
19. The apparatus as defined in claim 13, wherein the scheduling
period of the apparatus is the frame length of a physical channel.
Description
FIELD OF TECHNOLOGY
The present invention generally relates to the techniques
concerning downlink packet scheduling in a mobile communication
system. In particular, the present invention relates to a method
and apparatus for the optimum packet combined scheduling of
dedicated transport channels for packet services in UMTS (Universal
Mobile Telecommunications System) downlinks.
BACKGROUND ART
UMTS (Universal Mobile Telecommunications System) is the 3.sup.rd
generation mobile communication system of the radio technology
using WCDMA. In the system architecture of the UMTS terrestrial
radio access network (UTRAN) shown in FIG. 1, a radio network
controller (RNC) is connected to a core network via an Iu
interface, the RNCs are interconnected via an Iur interface, and
one RNC is connected to one or more Node Bs via an Iub interface. A
Node B contains one or more cells, the cell being a basic unit to
which a user equipment (UE) has wireless access (not shown),
wherein a radio interface between the UE and the UTRAN is a Uu
interface (not shown).
In the protocol documents of the standardization organization 3GPP
(the 3rd Generation Partnership Project) of the UMTS, there mainly
are TS25.2XX, TS25.3XX and other serial specifications relevant to
the UMTS radio interface protocol. In the UMTS radio interface
protocol architecture as shown in FIG. 2, the bottom layer is a
physical (PHY) layer, and in a control plane, above the physical
layer are a media access control (MAC) layer, a radio link control
(RLC) layer and a radio resource control (RRC) layer, respectively;
in a user plane, the radio interface protocol consists of the
physical layer, the MAC layer, the RLC layer and a packet data
convergence protocol (PDCP) layer, wherein the PDCP layer is only
for a packet-switch (PS) domain. Physical channels are provided by
the physical layer, logical channels are provided between the MAC
layer and the RLC layer, and transport channels are provided
between the MAC layer and the physical layer.
In the UMTS radio access network (UTRAN) of R99, the logical
channels of the control type include BCCH (Broadcasting Control
Channel), PCCH (Paging Control Channel), DCCH (Dedicated Control
Channel), CCCH (Common Control Channel), etc.; the logic channels
of the traffic type include DTCH (Dedicated Traffic Channel), CTCH
(Common Traffic Channel), etc. Uplink transport channels comprise
RACH (Random Access Channel), CPCH (Common Packet Channel), DCH
(Dedicated Channel), etc., while downlink transport channels
comprise BCH (Broadcast Channel), PCH (Paging Channel), FACH
(Forward Access Channel), DSCH (Downlink Shared Channel), and DCH,
etc. One of the primary functions of the MAC layer is to map
logical channel as transport channel. FIG. 3 shows the mapping
relations between downlink logic channels and transport
channels.
According to the 3GPP specification including 3GPP TS 25.212, 3GPP
TS 25.302 and other documents, a TFI (Transport Format Indication)
of each transport channel corresponds to one transport format in a
Transport Format Set (TFS) of the transport channel. In each TTI
(Transport Time Interval), as illustrated in FIG. 4, an upper layer
transports TBs (Transport Blocks) of the respective transport
channels to a PHY layer based on a certain Transport Format
Combination (TFC); the PHY layer then combines the TFI information
from the different transport channels into a TFCI (Transport Format
Combination Indication), encodes it and transports it on a TFCI
field of the PHY channel; after that, a receiving terminal decodes
the TFCI field so as to precisely receive the TBs from the
respective transport channels.
In UMTS, the main factor which influences downlink capacity,
coverage and other performances is a limited downlink power. The
total maximum transmit power of downlinks in a cell is determined
by a rated output power of a base station power amplifier, and this
power is typically divided into fixed static power, non-schedulable
power, and schedulable power, etc., as shown in FIG. 5. The fixed
static power is used for downlink common control channels in a
cell, such as common piloting, synchronizing, is paging and so
forth, and this fixed static power is determined by cell
configuration and other parameters; the non-schedulable power is a
power occupied by real-time services including conversational type
and streaming type services, wherein the real-time service allows
the selection of a certain rate, but the rate itself is still
required to be constant, so this power is non-schedulable; the
schedulable power is mainly used for NRT (Non Real-Time) packet
services such as interactive type and background type services,
which allows for a dynamic change in the rate, so the power
occupied by these services is schedulable.
When utilizing DCHs to transport interactive or background NRT
packet data, a total power of DCHs bearing packet services of a
plurality of different users forms the above non-schedulable power.
However, in the downlink direction of UTRAN, each UE has a DCH
functional entity MAC-d corresponding thereto, but the MAC-d
entities of different UEs are independent of each other. Therefore,
each cell must utilize a chief scheduling unit for conducting
combined packet scheduling of all the DCHs in the cell employed for
transporting NRT packet data. Nevertheless, the existing downlink
packet scheduling technology in the UMTS is mainly aimed at common
or shared channels such as FACH/DSCH, etc., and the scheduling of
the common or shared type channels can be basically summed up as a
problem that a plurality of input data streams shares an output
channel with limited bandwidth resources. In practice, as for the
familiar scheduling problems in this type of telecommunications,
there are a large number of mature and effective algorithms at
present, typically such as Round-Robin, WFQ (Weighted Fair
Queuing), WF.sup.2Q (Worst-case Fair Weighted Fair Queuing) and so
forth. Please refer to "H. Zhang, Service disciplines for
guaranteed performance service in packet-switching networks,
Proceedings of the IEEE, vol. 83, pp. 1374-1396, Oct. 1995", "V.
Bharghavan, S. Lu and T. Nandagopal, Fair queuing in wireless
networks: Issues and Approaches, IEEE Personal Communication, Vol.
6, No. 1, pp. 44-53, February 1999", and other documents.
SUMMARY OF THE INVENTION
The prior art lacks an effective and optimized method for combined
packet scheduling of DCHs used for packet services in the UMTS
downlinks. Thus, one object of the present invention is to provide
a method for downlink DCH combined packet scheduling and an
apparatus for realizing the same, and this method has a small
calculation amount, guarantees the fairness, priority and QoS
(Quality of Service) of different DCHs, and is optimized in the
sense of the maximum total data throughput.
A further object of the present invention is to solve the technical
problems of complex calculation, time-consuming and imperfectness
encountered during a programming of the downlink DCH combined
packet scheduling, and convert these problems into an optimization
problem with restrictions, thereby optimizing the downlink DCH
combined packet scheduling method.
The above objects of the present application are fulfilled by a
method for packet combined scheduling of dedicated transport
channels for packet services in UMTS downlinks, wherein dedicated
traffic channels (DTCHs) in logical channels are mapped as
dedicated channels (DCHs) in transport channels, and N DCHs, in
their respective input queues, queue up for being transported to
the corresponding M DPCHs, where M.ltoreq.N, the method being
characterized in that the method for packet combined scheduling of
the DCHs comprises the following steps: a) prior to each DCH
scheduling period, performing pre-selection processing of a
transport format combination of each DPCH according to the
predetermined restriction conditions for the DCH combined packet
scheduling, so as to determine a usable transport format
combination set TFCS.sub.m.sup.(2) for each DPCH; b) restricting a
total downlink transmit power
.times. ##EQU00001## of DCHs for NRT packet services to a
schedulable power not exceeding a schedulable power value in the
estimation of a total downlink power during said scheduling period,
i.e., the maximum allowable power value P.sub.k.sup.Scheduled,
where P.sub.k,n denotes an average transmit power required by the
N-th DCH in the k-th scheduling period, and P.sub.k.sup.Scheduled
denotes the maximum allowable power allocated to the DCHs which
bear NRT packet services in the estimation of downlink power in the
k-th scheduling period; c) based on the fairness of DCH
transportation and the QoS requirements of the DCH-borne services,
determining weighted values which the respective DCHs correspond to
in the optimization of the DCH combined packet scheduling; and d)
based on the results of steps a), b), and c), calculating the
maximum number of bit(s) which each DCH is schedulable to output,
using a 0-1 programming algorithm.
The above objects of the present application are also fulfilled by
an apparatus for performing packet combined scheduling of dedicated
transport channels for packet services in UMTS downlinks, wherein
dedicated traffic channels (DTCHs) in the logical channels are
mapped as dedicated channels (DCHs) in the transport channels, and
N DCHs, in their respective input queues, queue up for being
transported to the corresponding M DPCHs, where M.ltoreq.N; the
apparatus being characterized in that the apparatus for performing
the packet combined scheduling of the DCHs comprises: a
pre-selection processing unit for, prior to each DCH scheduling
period, performing pre-selection processing of a transport format
combination of each DPCH according to the predetermined restriction
conditions for the DCH combined packet scheduling, so as to
determine a usable transport format combination set
TFCS.sub.m.sup.(2) for each DPCH; a power restriction proportional
coefficient calculating unit for calculating a predicted/estimated
value .sub.,n of a proportional coefficient of an average transmit
power P.sub.k,n of the n-th DCH within the k-th scheduling period
and the number of bit(s) R.sub.k,n which the DCH is schedulable to
output within the scheduling period; a target function weighted
value calculating unit for, based on fairness of DCH transportation
and QoS requirements of the DCH-borne services, calculating
corresponding weighted values of respective DCHs in optimization of
the combined packet scheduling; and a 0-1 programming-based
optimization calculating unit for, based on output results of the
pre-selection processing unit, the power calculating unit and the
target function weighted value calculating unit, calculating the
maximum number of bit(s) which each corresponding DCH is
schedulable to output, by using a linear programming algorithm.
The method and apparatus for combined packet scheduling of the
downlink DCHs in the UMTS according to the present invention
realize the following goals: (a) maximizing the overall flow of
data output in each scheduling so as to effectively exploit radio
resources to the fullest extent, and meanwhile, supplying users
with the maximum data transport rate under the condition of limited
radio resources; (b) guaranteeing the fairness of the respective
DCH transportation and ensuring each user to obtain the maximum
data transport rate as much as possible; and (c) reflecting and
guaranteeing the priority requirements of the DCHs.
BRIEF DESCRIPTION OF THE ACCOMPANYING DRAWINGS
The modes for carrying out the invention are described below with
reference to examples and the accompanying drawings.
FIG. 1 is a schematic view showing the network architecture of the
UTRAN to which the present invention is applied;
FIG. 2 is a schematic view showing a radio interface protocol
architecture of the UTRAN to which the present invention is
applied;
FIG. 3 shows the mapping relations between downlink logic channels
and transport channels in the UTRAN to which the present invention
is applied;
FIG. 4 is a schematic view showing transport channels, transport
blocks and a transport format combination in the UTRAN to which the
present invention is applied;
FIG. 5 schematically shows the composition of a downlink transmit
power in the UTRAN;
FIG. 6 schematically shows a packet scheduling model from downlink
DCHs to DPCHs in the UTRAN;
FIG. 7 schematically shows the corresponding relations between the
downlink DCHs and the DPCHs in the UTRAN; and
FIG. 8 is a schematic view showing the apparatus for packet
combined scheduling of downlink DCHs according to the present
invention.
BEST MODE FOR CARRYING OUT THE INVENTION
The best modes for carrying out the present invention are described
below with reference to the accompanying drawings. Those skilled in
the art can better understand the technical solutions of the
present invention based on the more detailed description of the
technical solutions of the present invention with reference to the
accompanying drawings.
Referring to FIG. 6, it shows the packet scheduling model from
downlink DCHs to DPCHs in the UTRAN. In the model shown in FIG. 6,
N DCHs (marked as DCH#1, DCH#2, DCH#3, . . . DCH#N, respectively),
in their respective input queues, queue up for being output to the
corresponding M DPCHs (marked as DPCH#1 DPCH#2 DPCH#3, . . .
DPCH#M), the maximum allowable rate of each DPCH being determined
by its spreading factor, and one or more DCHs can be multiplexed on
one DPCH, so M.ltoreq.N. In the UMTS, the frame length of the
physical channels is 10 ms, while the TTI length may be 10 ms, 20
ms, 40 ms, and 80 ms. DCHs of different UEs have different TTIs, so
one scheduling is performed every 10 ms in the combined packet
scheduling of downlink DCHs, and the scheduled transport blocks of
each DCH are transmitted to the physical layer. The starting point
of the present invention for solving the combined packet scheduling
problem of downlink DCHs is to convert the problem into an
optimization problem with restrictions.
First of all, the combined packet scheduling of downlink DCHs has
three main goals: (a) maximizing the overall flow of data output in
each scheduling; (b) guaranteeing the fairness of the respective
DCH transportation; and (c) reflecting and guaranteeing the
priority requirements of the DCHs. Among these, (a) is the basic
requirement of packet scheduling, thereby effectively exploiting
radio resources to the fullest extent and supplying users with the
maximum data transport rate under the conditions of limited radio
resources on the other hand. (b) is another requirement of packet
scheduling, because the overall flow of data, instead of, the data
flow of each DCH is maximized in the first goal; in order to obtain
the maximum data transport rate from the angle of each user, the
fairness must be sufficiently ensured. Meanwhile, different DCHs
may have different priorities which are mainly determined by the
priorities and other factors of DCH-borne services. For example,
according to the 3GPP specification TS23.107, the QoS of the UMTS
interactive-type service has a "Traffic Handling Priority". The
packet scheduling mainly affects time delay, data rate and other
QoS performances of packet data services, so the combined packet
scheduling of downlink DCHs should reflect and guarantee the QoS
priority requirements of the DCHs. The priority resulting from the
different QoS requirement of a service is actually a static
priority. In addition, in each scheduling period, other factors may
also produce a dynamic priority, and a typical example is a status
PDU (Protocol Data Unit) packet in a RLC of an AM (Acknowledged
Mode). The RLC of the AM adopts the ARQ (Automatic Repeat Request)
technique and provides a low error-rate radio bearer capability at
the sacrifice of time delay. Since the NRT packet service is
error-sensitive rather than delay-sensitive, so the RLC of the AM
is usually employed. The status PDU is used to transmit ARQ
information from a receiving terminal of the RLC entity of the AM
to a transmitting terminal. Thus, in order to reduce delay, a
higher priority is needed for transmission.
On the other hand, the combined packet scheduling of downlink DCHs
should further satisfy the following restriction conditions: (a) in
each scheduling period (10 ms here), the total downlink transmit
power of all the DCHs for NRT packet services should not exceed a
schedulable power value within the estimation of the total downlink
power in this period; (b) in each scheduling period, the size and
number of the transport blocks which each DCH is scheduled to
output should meet the limit of a TFCS (Transport Format
Combination Set) which the DCH corresponds to; (c) a different DCH
has a different TTI, but the scheduling takes a radio frame as a
period. Thus, in one scheduling period, when a certain DCH has been
activated, namely, if the DCH is scheduled to have output a certain
amount of transport blocks in the previous scheduling period, the
DCH cannot be re-scheduled to output new transport blocks in the
current scheduling period but should conduct a new scheduling until
the next TTI of the DCH arrives; (d) in each scheduling period, the
number of transport blocks which a DCH is scheduled to output not
only depends on a scheduling system but also depends on the amount
of data waiting to be transported in an input queue. In order to
avoid generating unnecessary padding bits in a RLC layer which
causes waste of radio transport bandwidth, the transport capacity
which a transport format of the DCH corresponds to and which is
selected by the scheduling should be made less than the amount of
data to be transported in the input queue of the DCH in the current
scheduling period.
1. Target Function in DCH Combined Packet Scheduling
According to the foregoing goals of DCH combined packet scheduling,
the present invention adopts the following linear target function
to solve the optimal packet scheduling output:
.times..times. ##EQU00002##
where J.sub.k represents a target function of the k-th scheduling
period, w.sub.k,n represents a weighted value which the n-th DCH
corresponds to in the k-th scheduling period, and R.sub.k,n is the
to-be-solved number of bit(s) which the n-th DCH is scheduled to
output in the k-th scheduling period, whose value domain, as
described below, is a discrete limited area. As mentioned above,
the priorities resulting from different QoS requirements of a
service are a static priority, so the present invention defines the
priorities as a static weighted value portion of the above target
function. At the same time, in order to reflect the fairness
requirements of DCH transportation in the goals of the combined
packet scheduling, the present invention proposes to characterize
the fairness as a dynamic weighted value portion of the above
target function, that is,
w.sub.k,n=w.sub.n.sup.QoSw.sub.k,n.sup.Fair (2)
where w.sub.n.sup.QoS represents a static weighted value which
determined by the priority of a service borne by the n-th DCH and
which the n-th DCH corresponds to, w.sub.k,n.sup.Fair represents a
dynamic weighted value which the n-th DCH corresponds to and which
reflects the fairness requirements of the DCH in the k-th
scheduling period.
Need to particularly explain that the fairness has nothing to do
with the DCH rate, because the transport rate of a DCH is
determined by a bandwidth or spreading factor of a DPCH which the
DCH corresponds to, and both a DCH at a high rate and a DCH at a
low rate should acquire the same fair chance for a scheduled
output. Therefore, the present invention proposes, in order to
reflect the fairness requirements of each DCH transportation, to
use the historical (i.e. in the past period of time) scheduling
frequency of the DCH and a relative amount of each scheduling,
i.e., "scheduling index", to determine the above dynamic weighted
value w.sub.k,n.sup.Fair reflecting the DCH fairness requirements.
According to the present invention, the "scheduling index" of the
n-th DCH in the (k-l).sup.th scheduling period is defined as
.eta..times..times..noteq..times. ##EQU00003##
In Formula (3), R.sub.k-l,n represents the number of the bit(s)
which the n-th DCH is scheduled to output in the (k-l)-th
scheduling period, and max{R.sub.k-l,n} represents the maximum of
the schedulable bit which the n-th DCH corresponds to in a discrete
limited area in the (k-l)-th scheduling period. Obviously, the
value of .eta..sub.k-l,n is between 0 and 1, and since the value
domain of R.sub.k-l,n itself is a discrete limited area,
.eta..sub.k-l,n is also a discrete limited area. According to the
definition of the above "scheduling index", this parameter is
irrelevant with the rate of DCH.sub.n and it reflects the scheduled
degree of the DCH.sub.n in the (k-l)-th scheduling period.
By filtering the scheduling index of the n-th DCH during the past
period of time, we can obtain a historical average scheduling index
.eta..sub.k,n of the n-th DCH. According to the present invention,
one of the following three smoothing filtering ways can be
preferable employed:
.times..eta..times..times..eta..times..times..times..eta..times..times..-
lamda..times..eta..times..times..times..eta..beta..eta..beta..eta..times..-
times..times. ##EQU00004##
In the above formula, L represents the number of the past
scheduling periods participating in the smoothing filtering,
forgetting factor being .lamda..epsilon.(0, 1], and the function of
the factor lies in adjusting the significance of the past
scheduling index in the filtering. The closer to 0 the forgetting
factor is, the smaller function the past scheduling index performs,
and when I=1, Formula (4b) will be the average in the L-th period
as represented in Formula (4a). Formula (4c) actually indicates
that a first-order IIR (Infinite Impulse Response) is employed to
conduct the smoothing filtering, wherein the filter coefficient is
.beta..epsilon.[0, 1].
After solving the smoothing-filtered historical scheduling index
.eta..sub.k,n of DCH.sub.n using Formula (4), a dynamic weighted
value w.sub.k,n.sup.Fair which reflects the fairness requirements
of DCH.sub.n in the current scheduling period can be expressed,
according to the present invention, by a monofonic decent function
of .eta..sub.k,n. Preferably, one of the following two computing
methods can be used to determine the dynamic weighted value:
.eta..times..times..times..eta..times..times..times.
##EQU00005##
In Formula (5b), in order to prevent .eta..sub.k,n from being
excessively small, which may result in value instability, a
threshold value can be further used to limit the maximum weighted
value.
Finally, as for the dynamic priority resulting from ARQ and other
factors, the invention does not take it as a weighted value of the
target function, but adopts the following queue management
technique to reflect the dynamic priority: taking the status PDU
scheduling in a RLC of the AM as an example, once an input buffer
queue of any DCH receives packets containing the status PDU, it
immediately puts the packets at the forefront of the input buffer
queue of the DCH so as to enable the status PDU to get the
preferential transmission chance.
2. Restriction conditions of DCH Combined Packet Scheduling
Based on the foregoing restriction conditions of the DCH combined
packet scheduling, the total downlink transmit power of all the
DCHs for NRT packet services in each scheduling period should be
not more than the schedulable power value in the estimated total
downlink power in the same period. These restriction conditions can
be expressed as:
.times..ltoreq. ##EQU00006##
In the above formula, P.sub.k,n represents the average transmit
power required by the n-th DCH in the k-th scheduling period, and
P.sub.k.sup.Scheduled represents the maximum allowable power
allocated to DCHs bearing NRT packet services from the estimated
downlink power in the k-th scheduling period.
With the influences of rate matching being ignored, an average
transmit power in the downlink direction of a DCH within a radio
frame can be approximately expressed as a product of the bit rate
of the DCH and a proportional coefficient, wherein the bit rate of
the DCH corresponds to the number of the bit(s) the DCH within the
radio frame which is scheduled to be output in a scheduling period.
However due to propagation loss, multi-path fading and other
factors of a wireless channel, the proportional coefficient thereof
dynamically changes in each radio frame. Thus, the present
invention proposes to estimate the proportional coefficient of the
current scheduling period by using a linear prediction filter. For
instance, a first-order .alpha. tracker/predictor can be more
simply utilized:
c.sub.k,n=(1-.alpha.)c.sub.k-l,n+.alpha.c.sub.k-l,n (7)
In Formula (7), the filter coefficient is a .alpha..epsilon.[0, 1];
c.sub.k,n represents a predicted/estimated value of the
proportional coefficient of the average transmit power of the n-th
DCH and the number of the bit(s) the n-th DCH scheduled to be
output in the k-th scheduling period, and c.sub.k-l,n is a ratio of
a measurement of an actual transmit power of the n-th DCH and the
number of the bit(s) actually scheduled to be output in the (k-l)th
scheduling period. Thus, the power restriction conditions of DCH
combined packet scheduling in Formula (6) can be further expressed
as:
.times..times..ltoreq. ##EQU00007##
To satisfy other above-analyzed restriction conditions for DCH
combined packet scheduling, the present invention sets forth, prior
to each scheduling period, conducting a pre-selection process of
TFCS of each DPCH using these restriction conditions, the discrete
limited area of each DCH being determined by the pre-selected
usable TFCS of the corresponding DPCH.
(1) Restriction condition (b) (namely, in each scheduling period,
the size and number of TBs which each DCH is scheduled to output
should meet the limits of its corresponding TFCS) has actually
presented the TFCS of each DPCH, so this step is not necessary to
be executed in each scheduling period. Unless RRC is altered or
reallocated, TFCS of each DPCH will stay all the same in every
scheduling period. The usable TFCS of each DPCH obtained by Step
(1) is marked as TFCS.sub.m.sup.(0)={(TF.sub.1, TF.sub.2, . . . ,
TF.sub.S.sub.m)}.sub.m.sup.(0), where m=1, 2, . . . M, S.sub.m is
the TFC's dimention of the m-th DPCH, that is, S (m) DCHs are
multiplexed on the m-th DPCH;
(2) On the basis of restriction condition (c) (namely, in each
scheduling period, when a certain DCH has been activated, if this
DCH was scheduled to have output an amount of TBs in the previous
scheduling period, then the DCH, in the current scheduling period,
cannot be re-scheduled to output TBs but to conduct a new
scheduling until the next TTI thereof arrives), it is determined
whether or not each DCH in each DPCH has been activated in the
current scheduling period; if a certain DCH has been activated,
then the combination of all the transport formats excluding the
transport format being used by the DCH should be removed from the
usable TFCS obtained by Step (1) of the DCPCH which the DCH
corresponds to. The usable TFCS of each DPCH obtained by Step 2 is
marked as TFCS.sub.m.sup.(1)={(TF.sub.1, TF.sub.2, . . . ,
TF.sub.S.sub.m)}.sub.m.sup.(1), where m=1, 2, . . . M;
(3) On the basis of restriction condition (d) (namely, a transport
capacity which the transport format of the DCH selected by the
scheduling corresponds to should be not more than the amount of all
the data required to be transported in an input queue of the DCH
during the current scheduling period), from the usable TFCS of each
DPCH obtained by Step (2), all the TFCS which satisfy the following
conditions are further removed: the TFC should contain at least one
transport format indicating that the number of the bit(s)
transportable on the corresponding DCH within the scheduling period
is greater than the length of the current input buffer queue of the
corresponding DCH. The usable TFCS of each DPCH obtained by Step
(3) is marked as TFCS.sub.m.sup.(2)={(TF.sub.1, TF.sub.2, . . . ,
TF.sub.S.sub.m)}.sub.m.sup.(2), where m=1, 2, . . . M;
Thus, by utilizing the usable TFC set TFCS.sub.m.sup.(2) of each
DPCH obtained from the above TFCS pre-selection process, the
discrete limited area of each DCH will be determined. However, need
to note that when a plurality of DCHs are multiplexed on one DPCH,
the range spaces of all the DCHs of the DPCH obtained thereby are
coupled and they are limited by the limited combinations given by
TFCS.sub.m.sup.(2).
3. Converting DCH Combined Packet Scheduling Problem into 0-1
Programming Problem
Based on the foregoing analysis, in the k-th scheduling period
(subscript k is hereinafter omitted for clarity), the DCH combined
packet scheduling can be expressed as a mathematic programming
problem which adopts linear restrictions and linear target
functions in the discrete limited area.
.times..times. ##EQU00008##
.times..times..ltoreq. ##EQU00009##
Without limiting the value domain of the variables, the above
problem would be a typical linear programming problem. A quite
simple and effective computing method to solve the linear
programming problem is the simplex method which is widely used in
engineering. However, when the variables can only take their values
from the discrete limited area, no systematic and effective
computing method is available until now. The searching-type
algorithm is a basic method for solving such problem, but when the
number of the variables and the value domain thereof turn larger,
the calculation amount will become too enormous to be practically
used. Moreover, the coupling of the value domain space of DCHs
multiplexed on the same DPCH causes the problem to be solved more
complicatedly. Thus, the present invention utilizes the following
variable conversion technique to convert the aforesaid problem into
a special combination optimizing problem, i.e., the 0-1 programming
problem which can be easily solved.
To describe easily, as shown in FIG. 7, DCHs multiplexed to the
same DPCH are regarded as a groupe, so M DPCHs have M groups of
DCHs, each group containing S.sub.m (m=1, 2, . . . M) DCHs. The
DCHs are numbered as DCH.sub.m,1, DCH.sub.m,2, . . . ,
DCH.sub.m,S.sub.m based on their turns in the TFC of the
corresponding DPCH.sub.m, and R.sub.n, w.sub.n, c.sub.n in the
aforesaid optimizing problem are also correspondingly marked as
R.sub.m,j, w.sub.m,j, c.sub.m,j, where j=1, 2, . . . , S.sub.m.
Thus, the following equation is obtained:
.times. ##EQU00010##
Suppose the number of elements in the usable TFC set
(TFCS.sub.m.sup.(2)) of DPCH.sub.m is D.sub.m. Introduce D.sub.m
new 0-1 indicator variables q.sub.m,i.epsilon.{0,1}, i=1, 2, . . .
, D.sub.m to correspond to each element in the set, thereby
obtaining:
.times..function. ##EQU00011## and the following indication
restriction being satisfied:
.times. ##EQU00012##
In Formula (10), r.sub.m,j(i) is the number of transmissible bit(s)
in the current scheduling period indicated by the transport format
of the j-th DCH contained in the i-th TFC of the
TFCS.sub.m.sup.(2). Thus, by utilizing the newly introduced 0-1
indicator variables, we can not only remove the coupling of the
value domain space of DCHs multiplexed to the same DPCH, but also
convert the above optimizing problem into the following equivalent
0-1 programming problem:
.times. .times..times..gtoreq..times..di-elect cons..times..times.
##EQU00013##
.times..times. ##EQU00014##
The parameters W.sub.m,i and C.sub.m,i in the above formula can be
obtained by the following equations:
.times..function..times..times..times..times..function..times..times..ti-
mes. ##EQU00015##
Please note that in order to express the 0-1 programming problem in
a standard form, a slack variable S is introduced, and the target
function maximization is converted into the target function
minimization. Once the optimum solution of the aforesaid 0-1
programming problem is solved, the number of the bit(s) output in
the optimum scheduling of each DCH in the current scheduling period
can be solved from Formula (10).
4. Converting 0-1 Programming Problem into Linear Programming
Problem
There is no universal solving method for the 0-1 programming
problem at present, and most algorithms are proposed to solve some
particular types of problems. Note that the literature "A. K.
Khandani, Linear (Zero-one) programming approach to fixed-rate
entropy-coded vector quantisation, IEE Proceedings of
Communication, Vol. 146, No. 5, Oct. 1999, pp 275-282" presents a
solving method for some type of 0-1 programming problem. The 0-1
programming problem aims at the application of optimum bit
allocation in the fixed-rate entropy-coded vector quantization in
the above literature, but the mathematical model established based
on the 0-1 programming problem in this literature is identical with
the 0-1 programming of the DCH combined packet scheduling as
presented in the present invention. Therefore, the conclusions and
calculating methods for the 0-1 programming problem in this
literature can be directly applied to the present invention. The
whole text of this literature is cited here for reference.
According to the conclusion in Section 2.1 of the aforementioned
literature, the 0-1 restrictions over the indicator variables in
the 0-1 programming problem can be directly loosed, and the 0-1
programming problem is solved as a typical linear programming
problem. The optimum solution always guarantees that only at most
two indicator variables lie between 0 and 1, while other indicator
variables will be flat 0 or 1. As for the two variables between and
1, we can directly quantize them as 0 or 1 by the approximating
method, which exerts little and even negligible influence on the
optimum solution. In this way, based on relevant theories of the
linear programming and by adopting the effective simplex method,
the 0-1 programming problem will be solved through about
(M+1).about.2(M+1) times of iterations.
In addition to the universal linear programming calculation
methods, the aforementioned literature also presents a quick
solution with decreased calculation amount to solve the
corresponding linear programming problem, in light of the
particularity of the 0-1 programming problem. Please refer to
Section 2.2 of the literature for detailed algorithms.
5. Conclusions of DCH Combined Packet Scheduling Method
In the above description of downlink DCH combined packet scheduling
method of the present invention, the following two possible
practical cases have not been discussed: one is that different
DTCHs may be multiplexed to one DCH in the MAC layer; the other is
that DCHs for transporting NRT packet services and DCHs for
transporting RT services, such as the AMR (Adaptive Multi-Rate)
voice service may be multiplexed to one DPCH.
In the first case, the respective DTCHs multiplexed to one DCH have
the same QoS requirements and share the same transport bandwidth of
the DCH in a TDM (Time Division Multiplexing) way. As for this
case, the present invention sets forth to first adopt typical
scheduling algorithms, such as Round-Robin, WFQ or WF.sup.2Q, to
conduct DTCH scheduled allocation for the corresponding DCH, and
then to apply the method of the present invention to perform the
downlink DCH combined packet scheduling.
As for the second case, the present invention sets forth to
separately process the DCHs for transporting NRT packet services
and the DCHs for transporting RT services, namely, only considering
the influence of DCHs for transporting NRT packet services in the
power restrictions, and when determining the usable TFC set, only
adopting the descending dimensional TFC formed by the transport
formats of the DCHs for transporting NRT packet services, so that
the DCH combined packet scheduling method presented in the present
invention is still usable.
In summary, an apparatus for downlink DCH combined packet
scheduling presented according to the present invention is shown in
FIG. 8. The apparatus is made up of 4 functional units: TFCS
pre-selection processing Unit, power restriction proportional
coefficient calculating unit, target function weighted value
calculating unit, and 0-1 programming-based optimum packet
scheduling calculating unit. The apparatus operates once in each
scheduling period to obtain the scheduling output result of each
DCH in each scheduling period.
As shown in FIG. 8, during each scheduling period, in the TFCS
pre-selection processing unit, as for the DCH to which different
DTCHs are multiplexed, first execute a DTCH scheduled allocation
for the corresponding DCH by adopting typical scheduling algorithms
such as Round-Robin, WFQ or WF.sup.2Q; then, as for an input buffer
queue of each DCH, based on the dynamic priority requirements such
as status PDU in a RLC of the AM, put the high-priority packet at
the forefront of the input buffer queue of the corresponding DCH;
afterwards, conduct pre-selection in light of restriction
conditions such as TFCS of the corresponding DPCH, whether the DCH
is activated, and the length of the input buffer queue of the DCH,
and obtain the usable TFC set of each DPCH in the current
scheduling period; refer to the above for the specific processing
steps. Correspondingly, the TFCS pre-selection processing unit can
comprise one or more modules to perform the processes independently
or jointly, for example, comprising: a DTCH scheduling module which
executes a DTCH scheduled allocation of the DCH based on
Round-Robin, WFQ or WF.sup.2Q scheduling algorithm; a priority
queuing module which puts the high-priority packet, such as the
status PDU in the RLC of the AM, at the front of the input buffer
queue of the corresponding DCH; a TFCS pre-selection processing
Module which conducts the pre-selection processing based on a TFCS
of the DPCH, the activated DCH, and the length of the input queue
of the DCH, thereby to obtain the usable TFCS.sub.m.sup.(2) of each
DPCH in the current scheduling period.
During each scheduling period, in the power restriction
proportional coefficient calculating unit, first calculate the
ratio of a measurement of each DCH's actual transmit power to the
recorded actual number of the bit(s) scheduled to be output in a
scheduling period prior to the current scheduling period; then
utilize a linear prediction filter to estimate/predict a
proportional coefficient of each DCH in the current scheduling
period, wherein the linear prediction filter can typically utilize
the first-order .alpha. tracer/predictor shown in Formula (7).
During each scheduling period, in the target function weighted
value calculating unit, first calculate average scheduling indexes
based on Formula (3) and Formula (4) by using the recorded
historical scheduled output of each DCH; then based on Formula (5),
calculate a dynamic weighted value in the target function which
reflects fairness requirement of each DCH; on the other hand,
determine the static weighted value in the target function which
reflects the QoS priority of each DCH on the basis of priority
information of services borne by each DCH; finally multiply the
static weighted value and the dynamic weighted value, thereby
obtaining a weighted value of the original target function. The
target function weighted value calculating unit can also comprise
one or more modules to perform the processes independently or
jointly, for example including: a dynamic weighted value
calculating module which calculates the dynamic weighted value in
the target function according to the recorded scheduling result of
each DCH based on the fairness requirement of each DCH; a static
weighted value calculating module which determines the static
weighted value based on the priorities of services borne by each
DCH; and a multiplier which multiplies the dynamic weighted value
and the static weighted value thereby to obtain the weighted value
of the target function.
During each scheduling period, after obtaining the output results
of the TFCS pre-selection processing unit, power restriction
proportional coefficient calculating unit and the target function
weighed value calculating unit, the 0-1 programming-based optimum
packet scheduling calculating unit calculates the maximum number of
the bit(s) of each DCH. As scheduled to be output. According to an
embodiment of the present invention, the 0-1 programming-based
optimum packet scheduling calculating unit can also comprise one or
more modules to perform the processes independently or jointly, for
example including: a 0-1 programming parameter calculating module
which calculates, based on Formula (II), parameters W.sub.m,i and
C.sub.m,i in the 0-1 programming problem of downlink DCH combined
packet scheduling; a linear programming calculating module which
calculates the optimum solution vector of indicator variables in
the 0-1 programming problem of downlink DCH combined packet
scheduling by utilizing parameters W.sub.m,i, C.sub.m,i and the
schedulable power estimation in the current scheduling period,
wherein the schedulable power estimation in the current scheduling
period is provided by the downlink power allocating unit of a cell
radio resource management module; and a DCH scheduled output bit
calculating module which calculates the number of the bit(s) each
DCH scheduled to be output in the current scheduling period based
on Formula (10) by using the optimum solution vector of the
indicator variables output by the linear programming calculating
module.
The downlink DCH combined packet scheduling apparatus according to
the present invention as shown in FIG. 8, the TFCS pre-selection
processing unit, the power restriction proportional coefficient
calculating unit, and the target function weighted value
calculating unit can be realized by different hardware or software
in parallel, or realized by identical hardware or software
sequentially, wherein the paralleling method can better meet the
real-time requirements of the packet scheduling. In addition, the
linear programming calculating unit of the 0-1 programming problem
is a primary unit performing intensive calculation amount in the
present invention, and it can realize high-speed computation by
using special hardware or software units so as to meet the
real-time requirements of the downlink DCH combined packet
scheduling.
The method and apparatus for downlink DCH packet combined
scheduling in UMTS of the present invention are illustrated above
with reference to the accompanying drawings. The present invention
is not limited for the use in UMTS. As for other channels having
the similar channel structure, the principles of the present
invention are applicable as well. As known by those skilled in the
art, variable modifications and improvements can be made to the
present invention according to the principles thereof, without
departing from the scope of the claims annexed to the
description.
* * * * *